Cache coherency, clustering, and Kernel virtualization

Matthew Dillon dillon at apollo.backplane.com
Sat Sep 2 14:46:39 PDT 2006


:Please excuse my newbness --- but how does this differ from UML?
:
:Thanks,
:--TongKe

    I'm somewhat familiar with UML and from what I can tell they are fairly
    similar.  I do not know about UML's scaleability, however, and insofar
    as I can tell UML is a standalone effort.  What we are contemplating
    goes far beyond that.   Don't think of a virtualized kernel as a 
    standalone entity... think of it as an abstraction of a set of resources
    which can then be safely and securely donated to a named cluster.  It
    just so happens that it can ALSO be used as a standalone entity in a
    wide range of applications.

    'securely' in this context is from the point of view of the person
    donating the resources, not from the point of view of the cluster using
    the resources.

    Given the choice between coding the cluster kernel-wide verses 
    attempting to partition resources (like a jail) within a kernel to
    donate to a cluster, it is pretty clear to me that it is better to
    code the cluster kernel-wide and use kernel virtualization to
    partition resources.

						-Matt





More information about the Kernel mailing list